Download X-ray spectral measurements for tungsten

X-ray spectral measurements for tungsten-anode from 20 to 49 kVp
on a digital breast tomosynthesis system
Da Zhang, Xinhua Li, and Bob Liua)
Department of Radiology, Division of Diagnostic Imaging Physics, Massachusetts General Hospital,
Boston, Massachusetts 02114
(Received 23 February 2012; revised 19 April 2012; accepted for publication 4 May 2012;
published 29 May 2012)
Purpose: This paper presents new spectral measurements of a tungsten-target digital breast tomosynthesis (DBT) system, including spectra of 43–49 kVp.
Methods: Raw x-ray spectra of 20–49 kVp were directly measured from the tube port of a Selenia
Dimensions DBT system using a CdTe based spectrometer. Two configurations of collimation were
employed: one with two tungsten pinholes of 25 lm and 200 lm diameters, and the other with a single pinhole of 25 lm diameter, for acquiring spectra from the focal spot and from the focal spot as
well as its vicinity. Stripping correction was applied to the measured spectra to compensate distortions due to escape events. The measured spectra were compared with the existing mammographic
spectra of the TASMIP model in terms of photon fluence per exposure, spectral components, and
half-value layer (HVL). HVLs were calculated from the spectra with a numerical filtration of
0.7 mm aluminum and were compared against actual measurements on the DBT system using W/Al
(target-filter) combination, without paddle in the beam.
Results: The spectra from the double-pinhole configuration, in which the acceptance aperture
pointed right at the focal spot, were harder than the single-pinhole spectra which include both primary and off-focus radiation. HVL calculated from the single-pinhole setup agreed with the measured HVL within 0.04 mm aluminum, while the HVL values from the double-pinhole setup were
larger than the single-pinhole HVL by at most 0.1 mm aluminum. The spectra from single-pinhole
setup agreed well with the TASMIP mammographic spectra, and are more relevant for clinical
purpose.
Conclusions: The spectra data would be useful for future research on DBT system with tungsten targets.
C 2012 American Association of Physicists in Medicine. [http://dx.doi.org/10.1118/1.4719958]
V
Key words: x-ray spectrum, spectroscopy, digital breast tomosynthesis
I. INTRODUCTION
The knowledge of x-ray spectra is a cornerstone of many
medical imaging researches and developments. For breast
imaging, spectral data are useful for optimizing imaging
techniques,1–7 evaluating performance of imaging systems
and components,8–10 calculating dose to the breast,11–16 estimation of breast composition,17–19 as well as dual energy and
contrast enhanced imaging.20–22 The computer codes by
Boone et al.23,24 cover both the mammographic and general
x-ray energy range and have served the medical imaging
research community for more than a decade.
In traditional mammography, the tube voltage ranges from
18 to 42 kVp. Today, a wider kilovolt peak (kVp) range is
used in digital breast tomosynthesis (DBT). For example, Selenia Dimensions (Hologic, Inc., Bedford, MA) systems employ
tungsten targets and operate at 20–49 kVp. The TASMIP
model of Boone et al. in the mammographic energy range23
covers 20–42 kVp with tungsten-anode, and the model is based
on a Eureka tube (model Mam Rad 100, Eureka x-ray Tube).
Their TASMIP model in the general x-ray energy range24,25
covers an energy range of 30–140 kVp, but the spectra were
based on a general x-ray tube designed for nonmammographic
applications, and the data of 30–49 kVp were extrapolated
from the measurements of 50–140 kVp. Considering the possi3493
Med. Phys. 39 (6), June 2012
ble differences in operating range of kVp anode angle, intrinsic
filtration, and composition of the target and anode disk26
between the existing data23–25 and a modern DBT system,27
we believe there is a need to obtain new spectral data.
For the measurements of x-ray spectra in diagnostic
imaging, researchers often use high purity germanium
(HPGe) detectors24 and compound semiconductors such as
cadmium zinc telluride (CdZnTe)28,29 and cadmium telluride
(CdTe).30–32 As compared to HPGe detector, which is generally bulky due to the need for liquid nitrogen cryogenic
equipment, CdTe and CdZnTe detectors with small thermoelectric cooler are much more compact and practical in
clinical environment.33 Particularly for mammographic spectroscopy, the limited space between the fixed x-ray tube and
breast holder makes the compact spectrometers preferable.
While CdZnTe spectra are generally distorted by charge
trapping and need to be corrected.28,29 Miyajima30 showed
that the correction for the distortion due to charge trapping is
not necessary when using a CdTe detector in diagnostic
x-ray spectroscopy. In this study, we conducted x-ray spectral measurements with a CdTe spectrometer on a Selenia
Dimensions DBT system. The resultant x-ray spectra of
20–49 kVp, starting from 2.5 keV and with energy bins of
0.5 keV, are shared as supplemental data for the research
community.40
0094-2405/2012/39(6)/3493/8/$30.00
C 2012 Am. Assoc. Phys. Med.
V
3493
3494
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
TABLE I. Details of the detector of the spectrometer.
Material
Density
Dimension
Electrode at front
Electrode at end
Window material
Bias voltage
CdTe
5:85 g=cm3
3 3 1 mm3
200 nm thick Pt
1000 nm thick In
0.1 mm Be
500 V
II. MATERIALS AND METHODS
II.A. Spectrometer used in this study
We used an integrated x-ray and gamma ray spectrometer
(X-123CdTe, Amptek, Bedford, MA) in this study. The
spectrometer has a CdTe Schottky diode detector of
3 3 1 mm3 , which is mounted on a thermoelectric cooler
to reduce electronic noise. The spectrometer combines the
detector, preamplifier, digital pulse processor, and multichannel analyzer together in one compact package. We used
1024 channels in the data collection and conducted energy
calibration with the 13.95 and 59.54 keV peaks of 241 Am.
The energy resolution of the spectrometer, in terms of the
full width half maximum (FWHM) of 14.4-keV 57 Co peak,
is 530 eV. Table I listed more details of the detector.
II.B. Spectral measurements
We removed the collimation and filter assembly on the
tube of a Selenia Dimensions system and measured the raw
x-ray spectra which are more flexible to use. Removal of the
collimation assembly also eased the positioning of the spectrometer. The spectrometer was placed with its axis perpendicular to the breast support (Fig. 1), and a metal board was
inserted between the spectrometer and the breast support to
provide extension toward the chest direction and to protect
the imaging detector from excessive radiation. We used
tungsten pinholes to limit the photon rate impinging onto the
detector, to prevent spectral distortion due to pileup.29,34
Because pinholes were used to collimate the incident
beam, and because the axis of the spectrometer was perpendicular to the detector, the received photon rate was sensitive
to the position of the spectrometer on the breast holder. We
carefully positioned the spectrometer regarding the direction
of the incident x-ray: the spectrometer was aligned to the
center of the tube’s output window in the left–right direction,
with the help of a laser pointer (see the left of Fig. 1). With
its position fixed in the L–R direction, the spectrometer was
then moved step by step in the anterior-chest direction. The
photon rates at different positions in the anterior-chest were
compared, and thus, the “sweet spot” at which the highest
photon rate was received by the spectrometer could be
located. We used this method to align the axis of the
spectrometer-pinhole assembly to the central ray.
The x-ray tube is composed by a tungsten focal track26 on
a molybdenum anode disk (Fig. 1). The anode angle is 16 .
The kVp accuracy was verified by the vendor to be within
0.5 kVp from the nominal values. We performed two experiments with different configurations of pinhole collimators to
examine the spectra measured with different acceptance
apertures to the focal spot.
In the first setup, a single pinhole of diameter of 25 lm
was placed on top of a brass spacer that separated the pinhole
from the input window of the spectrometer. This setup opened
a large acceptance aperture that did not restrict the reception
of photons from both the focal spot and its vicinity.
In the second setup, two pinholes were used with the brass
spacer separating them. The pinhole on top of the spacer had
200 lm diameter, and the one below the spacer had 25 lm
diameter. Based on the distances from the focal spot to the
first and second pinholes (418 and 456 mm, respectively),
the diameter of the acceptance aperture projected onto the
FIG. 1. Experimental setup of the spectral measurements.
Medical Physics, Vol. 39, No. 6, June 2012
3494
3495
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
(horizontal) plane of the focal spot was approximately 2 mm.
The double-pinhole setup was sensitive to its positioning in
the anterior-chest direction and moving away from the sweet
spot by just 1–2 mm significantly reduced the incident photon
rate. The good alignment of the axis of the double-pinholespectrometer assembly with the tungsten focal track could
minimize the contribution of off-focus x-ray photons from the
molybdenum anode disk in the measured spectra.
Since the filter-collimation assembly was removed during
the spectral measurements, the raw spectra had much higher
photon rate than the spectra filtered by the 0.7 mm aluminum.
In this situation, the control of the incident photon rate
became an important issue for the spectral measurement. The
common approaches of increasing the source to detector distance and making extended exposure with low tube current
did not work in this study: the source to image-receptor distance (SID) of the system is fixed as 70 cm, and the tube is
designed without fluoroscopic mode and with high exposure
rate to alleviate motion artifacts. On the other hand, excessively reducing incident photon rate, for example, by using
even smaller pinhole sizes, not only increases the difficulty of
alignment but also requires much more repeated exposures to
achieve sufficient photon statistics—a heavy burden for the
tube. We tackled this difficulty by reducing the tube current to
10–60 mA (depending on kVp) and setting the exposure time
at the maximum value of 2.5 s, using a service software from
Hologic. In addition, the two configurations of pinholes were
carefully chosen to achieve a balance between the incident
photon rate and the number of repeated exposures. The dead
time of the spectrometer in most of the measurements was
around 8%–13% (with the maximum being 15.6%), which
was well below the limit level of dead time (30%) recommended by the manufacturer. Four to six repeated exposures
were made for each of 20–49 kVp in the double-pinhole
setup, depending on the photon rate for the individual spectrum, and the total detected counts ranged from 15.8 to
19.5 K. Three to five repeated exposures were made in the
single-pinhole setup for [21, 23, 25,…, 49] as well as 20, 30,
and 40 kVp, and the total detected counts ranged from 11.1 to
15.6 K. Due to the concern of damaging the tube, no more
repeated exposures were performed.
II.C. Postprocessing of measured spectra
The spectra output from the spectrometer were with 1024
channels covering up to 65 keV, and these raw measurements
were rebinned into 0.5-bin-width spectra for the ease of the
subsequent steps. The kVp of the raw spectra was determined
by a second order polynomial fitting at the tail end of the
spectrum, which includes energies up to 1 keV beyond the
nominal tube voltage. Counts of channels above the kilovolt
(peak) values (due to pile-up), although very insignificant
(1–2 per channel if there is any), were removed.
A few potential causes of spectral distortion exist in the
spectral measurements with CdTe/CdZnTe detectors: carrier
trapping, the transmission (penetrating the detector) of primary x-rays, and the escape of secondary x-rays.28,30,33
According to the results in Ref. 30, the distortion due to
Medical Physics, Vol. 39, No. 6, June 2012
3495
carrier trapping is negligible when a thin CdTe detector is
used in the diagnostic x-ray energy range. In traditional
mammographic energy range (<35 kVp), the distortion from
the escape events are negligible due to the high Z materials
of the detector.29 On the other hand, with the range of kilovolt (peak) spanned to 49 kVp, it is necessary to examine the
distortion due to the escape events.
The stripping method was used to correct the measured
spectra for the escape events (transmitted primary and
escaped secondary x-rays)28,32,35
Nd ðE0 Þ Nt ðE0 Þ ¼
Emax
X
RðE0 ; EÞNt ðEÞ
E¼E0 þ0:5
RðE0 ; E0 Þ
;
(1)
where Nt ðE0 Þ is the number of photons after correction at
energy E0 ; Nd ðE0 Þ is the number of photons in the measured
spectrum at energy E0 ; Emax is the maximum energy in the
measured spectrum; RðE0 ; EÞ is the monoenergetic response
function, which is the proportion of deposited energy in the
detector at E when a photon of energy E0 is received; and
RðE0 ; E0 Þ is the full-energy absorption peak efficiency.
The monoenergetic response functions were simulated
using GEANT4 (Ref. 36) low energy electromagnetic physics
package with photon and electron processes above 250 eV,
including photoelectric effect, Compton scattering, Rayleigh scattering, and secondary particles generated after
these events. Similar to the work in Ref. 30 and 35, carrier trapping and dead layer of the crystal were not taken
into account in the simulation, because they were insignificant in thin (1 mm) CdTe crystals working at high bias
voltages.30 The setup of the simulation is summarized in
Table II.
II.D. Comparison of HVL from measurements
and from spectra
Half-value layers (HVLs) were calculated from the measured spectra and from TASMIP spectra using the method of
Ref. 24 to examine the beam quality of the spectra presented in
this work. Filtration of 0.7 mm pure aluminum was applied
numerically to the raw spectra before HVL calculation. The
mass attenuation coefficients of aluminum and air were
obtained from the NIST XCOM database,37 and the mass
energy absorption coefficients of air were obtained by combining the data of NIST XCOM and NIST XAAMDI.38 Experimental measurements of HVL were also conducted on the
TABLE II. Setup of the Monte Carlo simulation for determining the response
function RðE0 ; EÞ.
Incident beam
Interactions
Escape events
Carrier trapping
Dead layer
Photon samples
Input x-ray energy
Pencil beam centering at the front of the detector
Photoelectronic, Compton, and Rayleigh
Fluorescent x-ray generation down to 250 eV
Not included
Not included
1 106 for each keV
2–125 keV, with 0.5 keV step
3496
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
Selenia Dimensions system, with the 0.7 mm aluminum and
without paddle in the beam. The aluminum sheets used were
made of type-1145 aluminum (Fluke Biomedical Al Set, Model
07-434). The reasons to compare HVL of the spectra with
0.7 mm aluminum added filtration were (1) the raw spectra contain large components below 10 keV, which may not be well
detected by the ion chamber used in the exposure measurements and (2) the large low energy components also make the
calculation of exposure and HVL from the spectra data more
prone to errors.
III. RESULTS
III.A. Comparison of selected spectra
The spectra measured with the single-pinhole and doublepinhole configurations, as well the spectra of Boone’s
TASMIP (Ref. 23) at selected tube voltages (20, 30, 40, and
49 kVp) were compared in Fig. 2.
3496
The spectra in Fig. 2 were corrected with the stripping
method described in Sec. II.C and were normalized to their respective total counts for comparison. For 49 kVp which is not
covered by the TASMIP mammographic spectral model,23
only the measured double- and single-pinhole spectra were
compared. It should be noted that although x-ray spectra of
43–49 kVp are available in TASMIP general x-ray spectral
model,24 these spectra are much harder than the measured
spectra, because they were from a diagnostic x-ray tube with
aluminum tube port and additional filtration of 0.5–1.0 mm
aluminum were applied in their measurements. For example,
the mean energy of the 49-kVp raw spectrum of TASMIP in
Ref. 24 is 30.4 keV, while the mean energy of the 49-kVp raw
spectrum measured with single pinhole is 17.3 keV.
III.B. Comparison of HVL
HVL obtained from different sources were compared at
different kVp in Fig. 3. For the TASMIP spectra, only 20–42
FIG. 2. Comparison of photon number spectra obtained from single-pinhole and double-pinhole setups and from TASMIP model (Ref. 23). Both single- and
double-pinhole spectra were corrected for escape events. All spectra were normalized to their own total counts. In (d), the TASMIP spectrum was not shown
because the kVp is not covered in TASMIP for mammographic applications.
Medical Physics, Vol. 39, No. 6, June 2012
3497
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
FIG. 3. Comparison of HVL calculated from the spectra and from direct
measurements using aluminum foils. Additional filtration of 0.7 mm aluminum was applied to all the raw spectra, and breast paddle was not included.
kVp were covered, as the spectra above 42 kVp in TASMIP
(Ref. 24) were too hard—those spectra were from a diagnostic x-ray tube with an aluminum tube port. HVL calculated
from the corrected single-pinhole and double-pinhole configurations were presented from 20 to 49 kVp. HVL from the
double-pinhole spectra were plotted between 20 and 46 kVp.
III.C. The influence of escape correction
The full-energy absorption peak efficiency RðE0 ; E0 Þ of the
spectrometer is shown in Fig. 4, and it is in good agreement
with the curve calculated by Miyajima30 with the LSCAT/
EGS4 code, and the one calculated by Bazalova and Verhaegen35 using EGSnrc/DOSXYZnrc code. The monoenergetic
response function RðE0 ; EÞ is also presented in Fig. 5, and the
effects of escaped Ka and Kb photons of Cd (23.2 and
26.1 keV) and Te (27.5 and 31 keV) can be seen from the four
bright lines to the right of the diagonal [RðE0 ; E0 Þ]. Note that
Fig. 4 corresponds to the diagonal of Fig. 5.
FIG. 4. Full-energy absorption efficiency of the 1 mm thick CdTe detector
calculated with GEANT4 Monte Carlo code.
Medical Physics, Vol. 39, No. 6, June 2012
3497
FIG. 5. Monochromatic response function RðE0 ; EÞ of the 1 mm thick CdTe
detector calculated with GEANT4 Monte Carlo code.
To study the influence of the escape correction on the
spectra, the spectra obtained at 49 kVp using the singlepinhole setup were compared before and after the stripping
correction in Fig. 6. The effect of escape correction on the
shape of the raw spectrum was small, due to the small portion of high energy components above the K-edges of Cd
(26.1 keV) and Te (27.5 keV) in the entire spectra. On the
other hand, when the spectra were numerically filtered by
0.7 mm Al, the effect of the escape correction became
clearer in Fig. 6: after correction, the spectrum became
slightly harder because the heights of energy bins below
26–27 keV were reduced and those above 26–27 keV were
increased. This was because the counts contributed by the
escape events were removed (stripped) from lower energy
bins and were added back to the corresponding high energy
bins [Eq. (1)].
To further study the influence of escape correction on spectra of different kVp, photon fluence per exposure (U=X)
(Refs. 24 and 39) was calculated for each kVp of the singlepinhole and TASMIP mammographic spectra, after being
numerically filtered by 0.7 mm aluminum (Fig. 7). Photon
FIG. 6. Spectra with and without the escape correction. The raw spectra
were obtained with the single-pinhole setup and were filtered numerically
with 0.7 mm Al.
3498
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
FIG. 7. Photon fluence per exposure calculated from the single-pinhole and
TASMIP spectra. All spectra were numerically filtered by 0.7 mm Al.
fluence per exposure is an important quantity in the calculation of detective quantum efficiency,24 and the details of calculating U=X can be found in Ref. 39. Note that the
separation between curves of the same collimation setup but
with/without escape correction becomes clearer beyond about
35 kVp: the relative difference is less than 0.8% below 35
kVp and is 1.2%–3.6% between 37 and 49 kVp.
IV. DISCUSSION AND CONCLUSION
In this work, we measured the raw x-ray spectra from a
Selenia Dimensions DBT system. The new spectral data
cover 20–49 kVp, while the existing data of tungsten target
for mammographic applications23 cover 20–42 kVp. The
raw spectra measured directly from the tube port without
added filtration can be used with numerical filtrations of any
material and thickness to match the actual settings of an
imaging application and are more flexible than spectra measured with added filtration.
Two configurations for collimating the incident x-ray
beam were employed: one with a narrow acceptance aperture achieved with a double-pinhole setup, and the other
with a wide acceptance aperture from on single pinhole.
We used these configurations to study the difference
between the spectra from the focal spot and the spectra
from the focal spot and its vicinity. The double-pinhole
spectra are generally harder than the single-pinhole ones, as
the tail parts toward the maximum kilo-electron-volt of the
double-pinhole spectra are higher than those of the singlepinhole ones. In addition, the contribution from the molybdenum anode disk of the target was smaller in the doublepinhole spectra. For example, at 49 kVp [Fig. 2(d)] where
the molybdenum peak at 17.5 keV is most significant, the
difference of this peak from its neighboring channels is
about 0.0044 for the single-pinhole setup, and is 0.0013 for
the double-pinhole setup.
Spectral distortions from different sources were considered in this study. From the work of Miyajima,30 carrier trapping is insignificant for a thin CdTe detector working at high
bias voltages. On the other hand, when the range of tube
Medical Physics, Vol. 39, No. 6, June 2012
3498
voltage spans to 49 kVp, whether the distortion due to
escape events is negligible needs to be studied. In Fig. 4, the
full-energy absorption efficiency is between 80% and 89%
for 32–49 keV. This means that 11%–20% of the incident
photons at these energies will be counted in lower energy
channels of the spectra, which necessitates the stripping correction. From Fig. 6, it can be seen that the corrected 49 kVp
spectrum begins to be slightly higher than the uncorrected
spectrum above 26–27 keV in the raw spectra, and this difference becomes clearer after both spectra were numerically
filtered by 0.7 mm Al. Note that 26–27 keV are near Cd’s
26.1 keV and Te’s 27.5 keV K-edges. The influence of
escape correction on spectra of different kVp can be seen in
Fig. 7: U=X of corrected spectra becomes higher than that of
uncorrected spectra beyond about 35 kVp, indicating harder
beam quality. This observation is consistent with the finding
of Ref. 30 that below 35 kVp the correction of escape events
is not necessary.
From Fig. 3, the difference in beam quality of the singlepinhole spectra and double-pinhole spectra can be observed
more easily. The HVL obtained with the double-pinhole
spectra is higher (e.g., 0.04–0.1 mm aluminum higher starting from 31 kVp) than that of single-pinhole spectra at each
kVp. The differences between the HVL of the doublepinhole spectra and the measured HVL are at most 0.14 mm
Al, while the HVL of the single-pinhole spectra agreed with
the measured HVL within 0.04 mm Al. It should be noted
that the HVL calculated in this study was based on mass
attenuation coefficients of pure aluminum, while in the
actual HVL measurements type-1145 aluminum sheets were
used (consisting of > 99:45% aluminum and other elements
of higher atomic numbers, e.g., Si and Fe), which could
cause smaller HVL measurements than if pure aluminum
sheets had been used. The reason for using the attenuation
coefficients of pure Al instead of type-1145 aluminum was
that the vendor only specified ranges of percentage elemental
composition of the alloy, and the exact composition of the
alloy may differ from batch to batch. The HVL calculated
using the corrected single-pinhole setup agreed with those
calculated with the TASMIP spectra within 0.025 mm Al. In
addition, the curves of photon fluence per exposure also
showed good agreement in Fig. 7. For the double-pinhole
spectra, we observed a drop of HVL values from 46 kVp
(0.87 mm Al) to 47 kVp (0.84 mm Al) and 48 kVp (0.85 mm
Al). This drop might be attributed to that the double-pinhole
setup (due to its small acceptance aperture) was sensitive to
the position of the focal spot which might have slight shift
after many exposures in the experiment or at the high end of
the kVp range.
The difference between the double- and single-pinhole
measurements lies in the fact that the double-pinhole has a
much smaller acceptance aperture (diameter of 2 mm projected
on the plane of the focal spot, see Sec. II.B) and what it measures are photons from the close vicinity of the focal spot. On
the other hand, the single-pinhole method has a much larger
acceptance aperture and it receives not only primary photons
but also off-focus radiation.26 In fact, the ion chamber used in
the HVL measurements also has a large acceptance aperture
3499
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
and that should be the reason why the measured HVL agreed
better with the single-pinhole spectra. For clinical purpose,
both the imaging subject and individual detector pixels have
large acceptance apertures, and the x-ray spectra they receive
should be close to the single-pinhole measurements.
In this work, we presented the new spectral measurements
on a Selenia Dimensions system, which has a tungsten target
and operates at 20–49 kVp. Despite the experimental difficulty and complexity, the spectra from the single-pinhole
setup showed good agreement with the existing TASMIP
mammographic spectra, and its beam quality is consistent
with the experimentally measured HVL. The comparison
between the spectra from the single-pinhole and doublepinhole setup revealed the difference between spectra from
the focal spot and spectra from focal spot and its vicinity,
which is of interest for future investigations on the impact to
image quality and patient dose. The new data and observations in this study should be useful for future research based
on DBT systems.
ACKNOWLEDGMENTS
The authors want to thank Dr. Baorui Ren of Hologic for
kindly providing the device and critical support for the
experiments and would like to acknowledge Amptek for
offering the spectrometer and technical support. This work
used the High Performance Clusters of the Partners Cooperation for Monte Carlo calculations, and the authors thank the
Enterprise Research IS group at Partners Healthcare for their
in-depth support.
a)
Author to whom correspondence should be addressed. Electronic mail:
[email protected]
1
A. Karellas and S. Vedantham, “Breast cancer imaging: A perspective for
the next decade,” Med. Phys. 35, 4878–4897 (2008).
2
G. Ullman, M. Sandborg, D. R. Dance, M. Yaffe, and G. A. Carlsson, “A
search for optimal x-ray spectra in iodine contrast media mammography,”
Phys. Med. Biol. 50, 3143–3152 (2005).
3
P. Bernhardt, T. Mertelmeier, and M. Hoheisel, “X-ray spectrum optimization of full-field digital mammography: Simulation and phantom study,”
Med. Phys. 33, 4337–4349 (2006).
4
N. T. Ranger, J. Y. Lo, and E. Samei, “A technique optimization protocol
and the potential for dose reduction in digital mammography,” Med. Phys.
37, 962–969 (2010).
5
A.-K. Carton, C. Ullberg, K. Lindman, R. Acciavatti, T. Francke, and
A. D. A. Maidment, “Optimization of a dual-energy contrast-enhanced
technique for a photon-counting digital breast tomosynthesis system: I. A
theoretical model,” Med. Phys. 37, 5896–5907 (2010).
6
A.-K. Carton, C. Ullberg, and A. D. A. Maidment, “Optimization of a
dual-energy contrast-enhanced technique for a photon-counting digital
breast tomosynthesis system: II. An experimental validation,” Med. Phys.
37, 5908–5913 (2010).
7
E. Samei and R. S. Saunders, Jr., “Dual-energy contrast-enhanced breast
tomosynthesis: Optimization of beam quality for dose and image quality,”
Phys. Med. Biol. 56, 6359–6378 (2011).
8
S. Vedantham, A. Karellas, and S. Suryanarayanan, “Attenuation
characteristics of fiberoptic plates for digital mammography and other
x-ray imaging applications,” J. X-Ray Sci. Technol. 11, 219–230
(2003).
9
J. Robert S. Saunders, E. Samei, J. L. Jesneck, and J. Y. Lo, “Physical
characterization of a prototype selenium-based full field digital mammography detector,” Med. Phys. 32, 588–599 (2005).
10
B. Lazzari, G. Belli, C. Gori, and M. R. D. Turco, “Physical characteristics
of five clinical systems for digital mammography,” Med. Phys. 34,
2730–2743 (2007).
Medical Physics, Vol. 39, No. 6, June 2012
11
3499
G. Richard Hammerstein, D. W. Miller, D. R. White, M. Ellen Masterson,
H. Q. Woodard, and J. S. Laughlin, “Absorbed radiation dose in
mammography,” Radiology 130, 485–491 (1979), http://radiology.
rsna.org/content/130/2/485.full.pdfþhtml.
12
X. Wu, G. T. Barnes, and D. M. Tucker, “Spectral dependence of glandular tissue dose in screen-film mammography,” Radiology 179, 143–148
(1991).
13
J. M. Boone, “Glandular breast dose for monoenergetic and high-energy
x-ray beams: Monte Carlo assessment,” Radiology 213, 23–37 (1999).
14
J. M. Boone, “Normalized glandular dose (DgN) coefficients for arbitrary
x-ray spectra in mammography: Computer-fit values of Monte Carlo
derived data,” Med. Phys. 29, 869–875 (2002).
15
I. Sechopoulos, S. Suryanarayanan, S. Vedantham, C. D’Orsi, and A.
Karellas, “Computation of the glandular radiation dose in digital tomosynthesis of the breast,” Med. Phys. 34, 221–232 (2007).
16
D. R. Dance, K. C. Young, and R. E. van Engen, “Estimation of mean
glandular dose for breast tomosynthesis: Factors for use with the UK,
European and IAEA breast dosimetry protocols,” Phys. Med. Biol. 56,
453–471 (2011).
17
J. Kaufhold, J. A. Thomas, J. W. Eberhard, C. E. Galbo, and D. E. G. Trotter, “A calibration approach to glandular tissue composition estimation in
digital mammography,” Med. Phys. 29, 1867–1880 (2002).
18
M. Yaffe, “Mammographic density. Measurement of mammographic
density,” Breast Cancer Res. 10, 209 (2008).
19
J. L. Ducote and S. Molloi, “Quantification of breast density with dual
energy mammography: An experimental feasibility study,” Med. Phys. 37,
793–801 (2010).
20
J. M. Lewin and L. Niklason, “Advanced applications of digital mammography: Tomosynthesis and contrast-enhanced digital mammography,”
Semin. Roentgenol. 42, 243–252 (2007).
21
S. C. Kappadath and C. C. Shaw, “Quantitative evaluation of dual-energy
digital mammography for calcification imaging,” Phys. Med. Biol. 49,
2563–2576 (2004).
22
A. D. Laidevant, S. Malkov, C. I. Flowers, K. Kerlikowske, and J. A.
Shepherd, “Compositional breast imaging using a dual-energy mammography protocol,” Med. Phys. 37, 164–174 (2010).
23
J. M. Boone, T. R. Fewell, and R. J. Jennings, “Molybdenum, rhodium,
and tungsten anode spectral models using interpolating polynomials with
application to mammography,” Med. Phys. 24, 1863–1874 (1997).
24
J. M. Boone and J. A. Seibert, “An accurate method for computergenerating tungsten anode x-ray spectra from 30 to 140 kv,” Med. Phys.
24, 1661–1670 (1997).
25
T. R. Fewell and R. E. Shuping, Handbook of Mammographic X-Ray
Spectra (HEW, Rockville, MD, 1978).
26
J. T. Bushberg, J. A. Seibert, E. M. Leidholdt, Jr., and J. M. Boone, The
Essential Physics of Medical Imaging, 2nd ed. (Lippincott Williams and
Wilkins, Philadelphia, 2002), pp. 97–144.
27
B. Ren, C. Ruth, J. Stein, A. Smith, I. Shaw, and Z. Jing, “Design and
performance of the prototype full field breast tomosynthesis system with
selenium based flat panel detector,” Proc. SPIE 5745, 550–561 (2005).
28
S. Miyajima, K. Imagawa, and M. Matsumoto, “CdZnTe detector in
diagnostic x-ray spectroscopy,” Med. Phys. 29, 1421–1429 (2002).
29
S. Miyajima and K. Imagawa, “CdZnTe detector in mammographic x-ray
spectroscopy,” Phys. Med. Biol. 47, 3959–3972 (2002).
30
S. Miyajima, “Thin CdTe detector in diagnostic x-ray spectroscopy,”
Med. Phys. 30, 771–777 (2003).
31
E. D. Castro, R. Pani, R. Pellegrini, and C. Bacci, “The use of cadmium
telluride detectors for the qualitative analysis of diagnostic x-ray spectra,”
Phys. Med. Biol. 29, 1117–11131 (1984).
32
K. Maeda, M. Matsumoto, and A. Taniguchi, “Compton-scattering measurement of diagnostic x-ray spectrum using high-resolution schottky cdte
detector,” Med. Phys. 32, 1542–1547 (2005).
33
R. Redus, J. Pantazis, T. Pantazis, A. Huber, and B. Cross,
“Characterization of CdTe detectors for quantitative x-ray spectroscopy,”
IEEE Trans. Nucl. Sci. 56, 2524–2532 (2009).
34
Handbook of X-Ray Spectrometry, 2nd ed., edited by R. E. V. Grieken and
A. A. Markowicz (Marcel Dekker, Inc., New York, NY, 2002).
35
M. Bazalova and F. Verhaegen, “Monte Carlo simulation of a computed
tomography x-ray tube,” Phys. Med. Biol. 52, 5945–5955 (2007).
36
S. Agostinelli et al., “GEANT4–A simulation toolkit,” Nucl. Instrum. Methods
Phys. Res. A 506, 250–303 (2003).
37
M. Berger, J. Hubbell, S. Seltzer, J. Chang, J. Coursey, R. Sukumar, D.
Zucker, and K. Olsen, “XCOM: Photon Cross Sections Database,”
3500
Zhang, Li, and Liu: 20–49 kVp x-ray spectra of tungsten target DBT
Radiation and Biomolecular Physics Division, PML, National Institute of
Standards and Technology (available URL: http://www.nist.gov/pml/data/
xcom/index.cfm). Last accessed December 9, 2011.
38
J. H. Hubbell and S. M. Seltzer, Tables of X-Ray Mass Attenuation
Coefficients and Mass Energy-Absorption Coefficients from 1 keV to
20 MeV for Elements Z ¼ 1 to 92 and 48 Additional Substances of Dosimetric Interest (National Institute of Standards and Technology, http://
www.nist.gov/pml/data/xraycoef/index.cfm, 2004).
Medical Physics, Vol. 39, No. 6, June 2012
39
3500
D. Zhang, X. Wu, M. Wong, Y. Ni, J. Rong, W. R. Chen, and H. Liu,
“Error analysis in the measurement of x-ray photon fluence: An analysis
on the uncertainty from energy calibration,” Proc. SPIE 7176, 71760I
(2009).
40
See supplementary material at http://dx.doi.org/10.1118/1.4719958 for the
spectral data measured with the single- and double-pinhole configurations
described in this paper. For more information on supplementary material,
see http://www.aip.org/pubservs/epaps.html.